[PVE-User] vzdump taking long time

[Patrick Westenberg]

Andreas Heinlein schrieb:
> Hello,
> 
> I am running nightly backups using vzdump on a single node Proxmox 5.1
> installation, backing up various VMs and LXC containers. I have one
> container (containing an Icinga2 installation) that has a backup size of
> 1.25 GB and takes almost 2 hours to back up, while another container
> running an elasticsearch db has a backup size of 12 GB and takes only 20
> minutes to back up.

I have similar experience with slow backup speeds and for me it feels
like the backup process is somehow speed limited.

INFO: status: 42% (9024045056/21474836480), sparse 18% (4036681728),
duration 418, 3/3 MB/s
...
INFO: status: 67% (14553120768/21474836480), sparse 39% (8453513216),
duration 632, 114/0 MB/s

INFO: transferred 21474 MB in 919 seconds (23 MB/s)

The backup storage is connected via multipathed 4 Gbps Fibre-Channel.
Disks on the storage server are 7200 rpm HDD as mirror pool (4x 2x 2 TB).

This storage used as storage space for VMs delivers more performance
than used as storage space. I also tried OCFS2 (current) and ZFS. No
difference. Setting vzdump --bwlimit has no effect.